The calculation of most probable numbers (mpn) was used for the estimation of numbers of infective entomopathogenic nematodes in soil. The mpn concept was first introduced in bacteriology as a means of estimating numbers of organisms in a substrate without a direct count, in cases where direct enumeration could not be applied. In the work reported here, soil samples infested with infective juveniles (IJs) of Heterorhabditis megidis (isolate HF85) were diluted with uninfested soil and the diluted soils were baited with mealworm larvae (Tenebrio molitor). The mpn of infective units of IJs in the undiluted soil was calculated. The mpn calculation was found to be applicable to entomopathogenic nematodes in soil under particular conditions. It could be successfully applied to data for the response of soil units but not to the data based on the response of individual insects, as the latter did not confirm to a Poisson series. The calculated mpn represented between 2.9 and 7.1% of the initial inoculum of IJs. It was suggested that IJs might act as a group in infecting an insect host. Using the data for Tenebrio mortality on parasitisation, the mpn based on quantal response of the mealworms would therefore not give the true density of IJs in the soil sample but the effective density, or the quantity of infective units. Although the biological significance of the infective unit needs further clarification, mpn was found to be a useful parameter for use in comparative experiments.